Woow...Thanks for all your information, you all really made my day. It's awesome to actually understand the physics of a real piano. So to wrap it up- is this a normal behaviour for a digital piano? Should I be concerned about the silenced note or is it just a Roland feature which is trying to simulate an acoustic piano?
Pianoteq (a "modelled" VST) does a pretty good job of simulating an acoustic piano's behavior:
. . . If you hit a key with "ff", followed by a series of "p",
. . . the original "ff" keeps sounding,
. . . and the "p" strikes add to the sound.
It sounds like a cymbal roll. Most DP's (e.g. my PX-350) don't understand how to do that.
To make it work well, you must use a 3-sensor action (which the PX-350 has), and _not_ completely lift the key, between strikes.
I don't think that's possible, with a two-sensor action.
With a 3-sensor action, you can generate several MIDI "note-on" events, without any intervening "Note off" events. You do that by not letting the key rise up all the way, but re-striking when it's halfway up.
It makes a difference, because the "note off" event is what tells the sound generator to lower the (virtual) damper and stop the sound.
If the F140 has a two-sensor action, it's not surprising that it should respond as you describe:
. . . When it gets the "note on / ff" event, it triggers the
. . . note at "ff" volume;
. . . when it gets the "note off" event, it drops the damper,
. . . . and the note decays;
. . . when it gets the "note on / p" event, it re-triggers the
. . . sample at "p" volume,
. . . and continues to decay the original note, even though
. . . the (virtual) damper is no longer on the (virtual) string.
If you look back, in the "Digital Pianos" forum, you'll find a
very long "DPBSD" (digital piano bull-[censored] detector) thread. It's a history of tests of DP's, and how well they simulate acoustic-piano behavior.
They've gotten better over the years, but they're still not perfect.